The Degree of Ethnic Bias in Kenya
Evidence suggests that ethnic divisions can have adverse effects on the economic and political stability of countries, including in Africa. Researchers in Kenya conducted multiple lab studies at the Busara Center for Behavioral Economics to explore the degree of ethnic bias in the local population and assess how bias may affect participants’ behavior. Contrary to existing assumptions about ethnic preferences in Kenya, researchers found no evidence of ethnic bias in decision-making related to sharing of economic benefits and political alignment.
Existing evidence suggests that ethnic divisions can have adverse effects on the economic and political stability of countries, including in Africa. However, there is disagreement among researchers on how ethnicity affects stability. Some theories suggest that individuals feel more generous towards those in their own ethnic group, referred to as “coethnics,” because of preferences or due to communication difficulties with other groups. Some researchers argue that urban centers in Africa, such as Nairobi, are especially likely to experience ethnic divisions because of social differences and high levels of job competition. To what extent do individuals in one African city, Nairobi, exhibit biased behavior due to their ethnic preferences?
Context of the evaluation
Kenya has long been associated with politically salient ethnic divisions. After gaining independence in the 1960s, Kenyan politics were dominated by the Kikuyu and Luo ethnic groups. In 2007, over 1,000 people died due to post-election violence, and hundreds of thousands more were displaced. The Luhya, Kamba, Kisii, and Kalenjins, together with the Kikuyu and Luo, comprised the six largest ethnic groups in Kenya in 2012.
This study was conducted in Nairobi at the Busara Center for Behavioral Economics, a research lab with facilities similar to those found in the US and Europe. Researchers partnered with local leaders to recruit participants from two low-income Nairobi neighborhoods, Viwandani and Kibera.
Details of the intervention
Researchers conducted scenario-based games at the Busara Center for Behavioral Economics to explore coethnic bias with 1,362 individuals.
Participants played three games using real money with anonymous partners, as well as with partners for whom age, education, and home region (a proxy for ethnicity) were known.
- Dictator Game: Participants were given KES 50 (US$0.60 in 2012), equivalent to an hour’s wage for a worker in Nairobi, and had to decide how much to share with an assigned partner.
- Public Good Game: Participants were given KES 60 (US$0.70 in 2012) and had to decide how much to keep for themselves versus how much to contribute to a group fund which was shared among the participant and two assigned partners.
- Choose-Your-Dictator Game: Participants were presented with two randomly drawn profiles, one of whom was a coethnic and the other a non-coethnic, and had to select who they would prefer to have play the role of Dictator in a Dictator Game.
Additionally, some participants were randomly selected to answer questions between games about ethnicity, political competition, and national identity. The questions were designed to remind participants about the prominence of ethnic identity in Kenya. Those in the comparison group were asked neutral questions.
Researchers conducted games from July to August 2012 and again from January to February 2013, immediately preceding the March 2013 national elections.
Results and policy lessons
Contrary to existing assumptions about ethnic preferences in Kenya, researchers found no evidence of bias based on ethnicity.
Games: In the Dictator Game, participants shared about one-third of their money regardless of their partners’ ethnicity. In the Public Good Game, participants shared about 46 percent of their money with coethnic partners, and an equal amount with mixed ethnicity groups. In the Choose-Your-Dictator Game, participants were equally likely to select coethnic and non-coethnic dictators. Asking participants questions about ethnicity, political competition, and national identity prior to game-playing had no effect on their behavior.
Timing of Games: Researchers found that overall levels of sharing fell for games played closest to the March 2013 national election. Dictator Game participants shared about 42 percent of their money in 2012, compared to 36 percent in the months immediately before the 2013 election. A smaller reduction was also seen in the Public Good Game. However, the researchers found no difference in ethnic bias during the games that were played during the lab sessions closer to and further from the election.
The lack of evidence for ethnic bias does not downplay the importance of ethnicity in Kenya, but suggests that ethnic divisions may be driven by factors other than ethnic bias alone.
Berge, Lars Ivar Oppedal, Kjetil Bjorvatn, Simon Galle, Edward Miguel, Daniel Posner, Bertil Tungodden, and Kelly Zhang. "How Strong are Ethnic Preferences?" NBER Working Paper No. 21715, November 2015.